Science News AI

Machine Learning Streamlines the Complexities of Making Better Proteins

Researchers at UC Berkeley and the Arc Institute unveiled MULTI‑evolve, a machine‑learning framework that predicts how multiple amino‑acid mutations affect protein function in a single experimental round. The workflow first estimates single‑mutation effects, then measures pairwise interactions, and finally trains a model to forecast higher‑order mutation outcomes. Tested on an autoimmune‑targeting antibody, a CRISPR editing protein, and another enzyme, MULTI‑evolve identified mutation combinations that outperformed the originals in laboratory assays. This approach promises to compress protein engineering cycles that traditionally require many iterative experiments.